Anthropological Linguistics VOLUME 44, NUMBER 4 WINTER2002 Offprint Relexification: A Reevaluation MICHEL DEGRAFF Contents Abstract 321 1 Whence ''relexification"? 321 2 "Strict relexification": The benefits of an explicit hypothesis. 325 2.1 The benefits. 325 2.2 The explicit hypothesis. 326 3 Reevaluation of the hypothesis. 328 3.1 The morphology: Matches and mismatches. 330 3.1.1 Is HC -yon a productive suffix (cf. French plantation)? . 331 3.1.2 On the structure of ordinal and cardinal numbers in HC, French and F::mgbe. 334 3.1.3 Inversive prefixes in Haitian Creole, French and {Fon)Gbe. 337 3.1.4 Suffixes for inhabitant names in Haitian Creole, French and F:mgbe. 343 3.1.5 The morphology: Recapitulation. .......... 345 3.2 A global comparison of lexica. .. 346 3.2.1 Empirical challenges. 346 3.2.2 Conceptual challenges. 347 3.3 Word order and functional heads: Empirical and theoretical challenges. 354 3.3.1 Theory-internal inconsistencies. ~5 3.3.2 Relexification and Haitian Creole functional heads. .. 359 3.3.3 The locus of word-order parameters. 363 4 Relexification vs. acquisition. 368 4.1 Relexification vs. second-language acquisition (L2A}. 369 4.2 Relexification vs. the sociohistory and demographics of colonial Haiti. 37 4 4.3 Relexification in L2A vs. the creation of"Creole" !-languages in first-Language acquisition (LlA). 382 4.4 LlA in Creole genesis. 386 4.5 An "L2A-L1A cascade" in Creole genesis ..... 391 Notes 394 References 405 ll Relexification: A Reevaluation MICHEL DEGRAFF Massachusetts Institute of Technology Abstract. According to one version of the Relexification Hypothesis, creole genesis is an instance of incomplete second-language acquisition whereby substrate speakers systematically fail to acquire the structural properties of their distant target, the superstrate Oexifier) language. The output of relexifica­ tion is an "early creole" with substrate-derived grammar and with superstrate­ derived phonetic strings. To date, the most thoroughly argued technical imple­ mentation of this version of the Reflexification Hypothesis is the study of Haitian Creole by Claire Lefebvre. In this article, I examine and refute the empirical-comparative, theoretical, and sociohistorical bases of the Reflexifica­ tion Hypothesis as implemented by Lefebvre and colleagues. Firstly, the basic assumptions and predictions of the Reflexification Hypothesis are inconsistent with well-documented details about the sociohistorical and linguistic profiles of Haitian Creole. Secondly, a systematic comparison of the morphosyntax of Haitian Creole with that of the languages which were in contact during its formation suggests a diachronic scenario that is fundamentally distinct from that envisaged in the Reflexification Hypothesis. Lastly, the foundational prin­ ciples and central claims of the Reflexification Hypothesis are mutually incon­ sistent, inconsistent with our current knowledge about language acquisition, or inconsistent with the Principles-and-Parameters framework in which Lefebvre couches her analyses. I conclude by sketching an alternative scenario for the diachrony of Haitian Creole that is compatible with the sociohistorical and lin­ guistic details of the language and with basic results in language-acquisition research. 1. Whence "relexification"? The Relexification Hypothesis is, alongside Bickerton's (e.g., 1984, 1999) Language Bioprogram Hypothesis, the most discussed scenario of creole genesis. One even finds relexification cited as unquestioned truism in the pages of widely read and sophisticated literary journals such as the Times Literary Supplement: "A creole [is] an African lan­ guage, or languages, upon whose syntax the vocabulary of another language is laid. The clearest example from the New World is [Haitian Creole], a creole which appears to be composed of French vocabulary overlaid on the grammar of the African language Ewe" (Greppin 2002:3-4). The most technical and most explicit relexification-based scenario for creole genesis is the one elaborated by Claire Lefebvre in her book Creole Genesis and the Acquisition of Grammar: The Case of Haitian (1998, hereafter "L "). In that scenario, creole creators in Saint-Domingue (modern-day Haiti) were essentially adult Africans during the years 1680-1740. "These adults were native speakers 321 322 ANTHROPOLOGICAL LINGUISTICS 44 NO.4 of languages of the Niger-Congo group and more specifically the Kwa languages, with a majority of Gbe speakers" (L:57). According to Lefebvre, these Niger­ Congo speakers were trying to acquire, with little success, the European target language, namely, French, to which they "had very limited access" (L:36; also see L:65, 386, 394). In this hypothesis, creole genesis is induced by repeated and across-the-board instances of an extraordinary sort of imperfect second­ language acquisition with Niger-Congo languages (the "substrate" languages) as native (first) languages and with French (the "superstrate" or "lexifier" language) as target (second) language. What makes these hypothetical creole-inducing instances of second­ language acquisition extraordinary is that they produce, as the structural endstate of reflexification, a nonnative idiolect with target-derived phonetics and a grammar that is virtually isomorphic to that of the learner's first lan­ guage-a grammar left mostly untouched by the target language. Extraor­ dinarily, whatever second-language acquisition took place in colonial Haiti during the initial creole-genesis period, it did not, according to Lefebvre, endow African learners with any substantial set of grammatical properties with ana­ logues in the European target grammar. In other words, the Relexification Hypothesis in the case of Haitian Creole is the story of how African slaves in colonial Haiti systematically failed to acquire any structural aspect of the French varieties spoken by the socially dominant classes. And, by definition, child learners (i.e., first-language acquisition) contribute nothing to the central stages of Creole genesis: relexification, which plays "a central role in creole genesis" (L:15), is the exclusive province of "adult native speakers in possession of mature lexicons and grammars" (L:394; also see L:10). It is the relexified interlanguages of individual African speakers from a variety of Niger-Congo languages that, by hypothesis, constitute the "early creole" grammars. Lefebvre's approach is mentalist: the early creole varieties, which are the main focus of the Relexification Hypothesis, are treated as mental objects, namely, as Internal Languages (!-languages) in Chomsky's (1986) sense. In other words, the early creole grammars (i.e., the postulated output of relexification) are, in Chomsky's terms, "internally represented in the mind/ brain" (Chomsky 1986:22) of individual African speakers. In Lefebvre's sce­ nario, each of these relexified and nonnative grammars will, by hypothesis, be structurally similar to the speaker's native I -language. In addition to relexification proper, Lefebvre also postulates two important processes: "reanalysis" and "dialect leveling." The output of the early creole grammars are subject to reanalysis, the technical definition of which, to be discussed below, is reminiscent of grammaticalization a la Bopp, Meillet, and others. Then dialect leveling takes place across the typologically disparate lects of early creole speakers, toward the creation of a relatively homogeneous creole grammar, which in turn is used as an identity marker for the creole community. Lefebvre's scenerio will be presented in more detail in section 2. My own 2002 MICHEL DEGRAFF 323 contention in this critique is that, given well-documented sociohistorical and linguistic evidence, this scenario cannot account for the diachrony and syn­ chrony of Haitian Creole, my native language. I will argue instead, in sections 3 and 4, in uniformitarian fashion, that the African adults who participated in the development of Haitian Creole, like other adult learners in situations of lan­ guage contact, did approximate various aspects of their target grammar, with a certain degree of restructuring influenced by, inter alia, the learners' first languages. Thus, alongside substrata! features, structural aspects of the super­ strate language did play a role in creole diachrony-a much larger role than Lefebvre's scenario allows for. In this vein, I will also argue that speakers of the earliest (Proto-)Haitian Creole varieties did analyze, or reanalyze, substantial amounts of morphosyntactic target (superstrate) patterns: they incorporated such target-oriented approximations, alongside other sorts of structural innova­ tions, into the early creole. Then, in section 4, I will argue, contra Lefebvre, that first-language acquisition by children, in this case by locally born (i.e., "creole") children, plays an important function in creole genesis in the sense of creation of stable and relatively homogeneous "Haitian Creole" !-languages. In fact, my argument will be even stronger: given the very theoretical premises of Lefebvre's analytical framework, it is logically impossible for relexification to play the "central role" that is envisaged by Lefebvre vis-a-vis the development of an aggregate of creole !-languages such as those of Haitian speakers. But I am already getting ahead of the story. So let me step back a bit and spell out the details of Lefebvre's scenario and some of its many positive contributions, before putting on my skeptical--devil's advocate's-hat for a constructive critique. Relexification,